A new genomic approach to describe the natural history of node negative breast cancer patients

Abstract

10058 Background: Breast cancer is a heterogeneous and complex disease beyond good/bad prognosis groups. Previous studies identified prognostic tools that are undergoing inter-group validation. Our present study aims at better defining pre-malignant vs. malignant lesions and tumors that will give local vs. distant metastases (DM) or early vs. late DM. Methods: We selected 3 groups of patients (pts) in a consecutive series of node-negative breast cancers with a very long follow-up (>10 years (y)) and available frozen tumor samples. Pts who did not receive any adjuvant treatment were selected: 60 with no relapse after 10 y, 29 with early DM before 5 y and 18 with late DM after 5y. Genomic profiling on 44K Agilent pangenomic arrays taking as reference non-relapsing pts (after 10y) or normal breast (Clonetech) allowed us to select specific genes related to DM occurrence or tumorigenesis. Random permutations were performed to assess the statistical significance of our prediction accuracy. Results: 1- we identified a 141 gene-profile and related genes distinguishing tumors with early DM from tumors with late DM (71% prediction accuracy). 2- Similarly, we identified a 285-gene signature for late DM vs. no relapse at 10 y (80% prediction accuracy). Genes were further classified according to family clusters allowing description of genes involved in DM occurrence. 3- a total of 435 genes were also significantly over-expressed in at least 90% of 148 tumours (compared to normal breast). Among those, a selection of the most significant genes were analysed for RT-PC expression throughout the various steps of tumour development (normal breast, benign tumors, in situ carcinomas, invasive carcinomas). Among the 15 most upregulated genes listed, 3 unknown genes appear, that are overexpressed in 98% of the tumors with an average fold change > 11 and are currently explored. Discussion: In this unique series of untreated node-negative breast cancers, identification of molecular profiles of early and late DM could be useful in better early prediction of breast cancer outcome. Based on genomic analysis, a molecular gene based classification of progression from benign tissue to aggressive tumors allows to dissect pathways towards malignancy and to identify early diagnostic markers or targets for prevention. No significant financial relationships to disclose.